Projects
A selection of projects that reflect my range as an engineer, from full-stack SaaS products to machine learning pipelines and cloud deployments.
PreLegal: AI Powered Legal Document Drafting Link to heading
Technologies: Python · FastAPI · Next.js · Docker · Claude (via OpenRouter) · GitHub Actions · Render
A full-stack SaaS platform that uses conversational AI to draft professional legal agreements in minutes. Users describe the agreement they need in plain English, the AI asks the right questions, and a clean legal document builds in real time. The platform supports 11 document types, including NDAs, Data Processing Agreements, and HIPAA Business Associate Agreements, based on Common Paper open standards vetted by 40+ attorneys.
Built with FastAPI and Next.js, with Claude via OpenRouter as the conversational drafting engine. Includes real user authentication, document persistence, PDF export, and a fully automated CI/CD pipeline deploying to Render.
Loss Ratio Prediction for Auto Insurance Link to heading
Technologies: Python · Pandas · Scikit-Learn · Docker · GCP
A machine learning project using an auto insurance dataset to predict loss ratios for insurance policy portfolios. Accurate loss ratio forecasting is foundational to insurance pricing and risk assessment. Poor predictions lead to underpriced risk and financial exposure.
The project followed the CRISP-DM methodology through the full lifecycle, from business understanding to model evaluation. Included data preparation, exploratory analysis, feature engineering, and regression modeling to build a reliable prediction pipeline. Infrastructure deployed using Docker and Google Cloud Platform.
(Private repository)
Housing Price Prediction Service Link to heading
Technologies: Python · Google Cloud Run · Scikit-Learn · Docker · GitHub Actions
A machine learning model deployed as a service to predict housing prices based on location, square footage, and property features. Built and trained a regression model, then deployed it to Google Cloud Run with automated CI/CD pipelines, enabling real-time predictions for real estate analysis.
Bank Marketing Analysis Link to heading
Technologies: Python · Pandas · Scikit-Learn
A data analysis project examining predictions on a bank marketing dataset, with a focus on identifying potential bias in model outputs across customer segments. Performed data preprocessing, exploratory analysis, and fairness metric computation to surface areas where the model produced inequitable outcomes, providing actionable insight for model improvement.
(Private repository)
If you have questions or want to explore a collaboration, reach out via my Contact page.